add cube2 env

rlcube/cube.blend

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:297bfe1ba114dd5ee6a1bfe2e6f8e73279d77c9e16cdf02e222036fdecabe8db
+oid sha256:a78d2ba0b0d1bbf77427eacdad672d7bc5ecdc95a5b4ffa5380470666af23924
 size 1963196

rlcube/rlcube/cube2.py

@@ -11,5 +11,5 @@ def train():
         print("--------------------------------")
         if terminated or truncated:
             break
-
+    print(env._is_solved())
     env.close()

rlcube/rlcube/envs/cube2.py

@@ -154,4 +154,10 @@ class Cube2(gym.Env):
             new_state[FRONT, 1, 1] = self.state[RIGHT, 1, 1]
 
         self.state = new_state
-        return self.state.copy(), 0, False, self.step_count >= 100, {}
+        return self.state.copy(), 1 if self._is_solved() else -1, self._is_solved(), self.step_count >= 100, {}
+    
+    def _is_solved(self):
+        for i in range(6):
+            if np.mean(self.state[i]) != self.state[i][0][0]:
+                return False
+        return True